Unsolved
This post is more than 5 years old
5 Posts
0
1774
July 8th, 2016 08:00
ECS Multi-Node Fails to start
Used the most recent master from github.
4 Cent Machines.
Install times out waiting for Auth.
It would appear that the certificates are missing.
glus-4:/ # ls -l /opt/storageos/conf/storageos.key /opt/storageos/conf/storageos.crt
ls: cannot access /opt/storageos/conf/storageos.key: No such file or directory
ls: cannot access /opt/storageos/conf/storageos.crt: No such file or directory
glus-4:/ #
This is the case on all 4 nodes.
Please help.
-Wade
No Events found!
wadeholler
5 Posts
0
July 8th, 2016 08:00
Looking at the nginx wrapper. This is the command that is hanging:
/opt/storageos/bin/zkutils getvnestkeyandcert
.....
Suggestions???
wadeholler
5 Posts
0
July 8th, 2016 11:00
KVM under OpenStack. I will try the use-urandom option.
Thank you for taking the time to reply
Wade
JasonCwik
281 Posts
0
July 8th, 2016 11:00
Hi Wade,
What virtualization platform are you using? We've seen issues with lack of entropy on some platforms that stall the generation of the SSL certificates. Nginx will wait to start until the SSL certs are generated. We added the --use-urandom option to the step1 script to substitute /dev/urandom for /dev/random inside the container so the certificates can be generated quickly.
JasonCwik
281 Posts
0
July 8th, 2016 14:00
Thanks Wade,
KVM/Openstack is the same platform others have seen the same issue with.